Click below to share this pageTweet
Published: May 17, 2016 11:40:00 AM
Published: May 15, 2016 9:56:00 AM
Published: Apr 21, 2016 10:20:47 AM
- Grand Challenges 2021: The Next Five Years
- Grand Challenges UCL 2034 Grants
- UCL Researchers: Why contribute to The Conversation?
- Grand Challenges Student Fund: up to £750 available for student led projects – More
Winning Project: RecommendME! : a personalised activity recommendation engine for the elderly
£10,000 Ageing Research Prize Workshop (12-13 December 2013):
Keeping people active, independent and well in later life
Project Team (left to right)
Alexandru Matei (UCL Computer Science)
Maryam Atakhorrami (Translational Research Office, School of Life and Medical Sciences, UCL)
Tarek Ahmed (UCL Eastman Dental Institute)
David Greenberg (UCL Ear Institute) - Project Lead
Navaz Davoudian (Bartlett School of Graduate Studies)
Ageing is a dynamic process and a journey that every person will experience differently. The physical and social changes that occur due to ageing can result in the reduced health and wellbeing of an elderly person as well as changes in their interaction with the wider society. Being active, which can be defined in both a social and physical context, is essential for a healthy, ageing society. Individuals who are transitioning from full-time employment to retirement have changing physical and mental needs, needs that we are only beginning to understand with the steady increase in life expectancy. The process of ageing and how it influences social and physical activity differs between individuals but these differences are often built upon psychological barriers of what the individual can, and cannot do. The consequences of ageing within a psychological framework of ‘cannot’, will inevitably result in a rapid transition from active to sedentary.
RecommendME! is envisioned as a digital platform for use by an ageing population, able to connect a user to relevant services, activities and peer groups. Personalised recommendations will be developed for each user based on their personal physical, mental and social capacity, their interests, geographical location and the activities undertaken by their peer group. By collecting and analysing trends in the activity data of individuals, a profile of abilities and relevant activities can be developed for the user-base. RecommendME! will enable the elderly to be more active by removing some of the psychological barriers in a tangible manner, motivating the user to be more active.
RecommendME! will be able to connect individuals to their peers via clusters of users with similar profiles, allowing individuals to attend activities together thus reinforcing the motivation to stay active. Through this manner of assistance RecommendME! will also have the capacity to identify gaps in voluntary and commercial services creating a constructive feedback loop.
· To create a platform that enables the elderly to be more active and to remain as active members of society.
· To develop a platform that is usable and functional for the target user-base
· To ensure that the platform maximises the potential of the user through recommendations based on individual ability.
· To improve upon existing levels of wellbeing in later life.
· Conduct research on the data required for a successful recommendation engine.
· Establish links with Camden Council to provide guidance on activities and to provide a link to elderly users for beta-testing of platform.
· Design questionnaires to discover pressing needs and preferences from the elderly.
· Work with data scientists from the ground up in building the platform to ensure data quality.
· Work with human-computer interaction experts to identify key attributes for the user interface.
· Produce a working recommendation platform and trial with 100 users.
Methodology 1) Behaviour-Change
The methodology will follow the general aim of encouraging individuals to remain physically and mentally active after retirement. A study in behaviour change will be conducted that is based on the three conditions of Capability, Opportunity and Motivation (the COM-B behaviour system). A combined qualitative and quantitative approach will be employed to assess physical and psychological capabilities of this target group as well as automatic and reflective motivational factors. This will allow us to enhance the pertinence of the social and physical opportunities recommended through the designed interface.
The methods will involve a set of questionnaires targeting the aforementioned group in the UK. The questionnaire study will be followed by interviews with groups of individuals in our target group in the borough of Camden, London. The combined results will provide the key physical and mental elements as well as motivational factors to be included in the input and output of the platform.
Methodology 2) Software
In order to support our objectives, a software infrastructure will be developed. There are two components that will be considered:
A social platform that serves several purposes. Firstly, this is the place where people can register and create their profiles. They can browse existing activities and get personalised recommendations. Some form of user to user interaction will also be supported. Secondly, this is the place for activity providers to create and promote their activities, by tagging them appropriately. Thirdly, the platform will provide a mechanism for users to subscribe to the activities they are willing to participate in.
The platform will be built following an agile software development methodology, since user feedback and user acceptance are very important. Also, given the budget constraints, an agile methodology will ensure we will get a working product within the timeframe. While development will be outsourced, we are relying on internal resources for project management and testing. Considering the time constraints, we will be looking into leveraging existing modules and libraries for building social platforms, open source or not. The recommendation engine will be developed in house. It will be custom built for this project, such that we increase the quality of the predictions. The engine will be initially trained using data from existing research and will continually improve while the system is in use.
The RecommendME! project will use the available £10k funding to build an investable and attractive proposal for a social enterprise venture. £7k has been earmarked to pay for the time required to develop and build the working demo including IT platform, machine learning unit and user interfaces. The remaining £3k has been allocated to facilitate with conducting focus groups and interviews that will guide the development of the platform.
(12 months) - envisaged start /end dates, key milestones, envisaged outputs
In months 1-3 a heavily researched business plan will be developed including detailed milestones, team development, data requirements and platform requirements. Months 3-6 will be spent building the digital platform and software infrastructure leading to the recruitment of 100 beta-testers in months 6-8 as well as initiating the development of an updated business plan based on insights gained as to the direction and development of the platform and any necessary further funding requirements. Running in parallel to the business development will be the development of the digital platform and literature research that will inform the interview questionnaire design. Throughout the whole 12 months there will be close interactions between the teams behind these two key components. Months 8-12 will be spent training and refining the platform, analysing interview and questionnaire data and platform based data collected from the beta-tester cohort. During this time applications will be finalised and submitted to relevant social enterprise investment grants such as Nesta (http://www.nesta.org.uk/) and UnLtd (http://unltd.org.uk/ ). The findings of the beta-tested platform with 100 users in Camden will be central to the business plan to secure further investment in the project.
Project completion date: January 2015
Background to this Research Prize Workshop
Page last modified on 14 feb 14 16:43